Automatic bidding, dear reader, is here to stay whether you hugged it with your arms wide open or neglected like the Dursleys neglected Harry P.
I started in this industry around the time automation was just getting started, let alone seen as a viable strategy for improved performance!
The ECPC (Enhanced Cost Per Click) bid was the first automated bid strategy I remember seeing Google debuting. In hindsight, this was a great way to get us Digital Strategists / Account Managers to start slowly, very slowly, and get used to automation.
Why the slow adoption?
As with most things, there are innovators, early adopters, and late adopters. I have often believed that being in the innovator / early adopter group is most beneficial when there is no reason other than to have some sort of first mover advantage. I think this is less the case with automated bids than with RSAs. However, the general adoption curve still applies. And that's the thing about the adoption curve – sooner or later we will all adopt the new technology / process. Automated bidding is no exception.
I think the only thing most of us struggled with when introducing automation was giving up control. Accepting automation meant letting go of much of what we were taught / at which we had previously achieved success. One of my colleagues, Dani Gonzales, explains this idea in greater detail on her blog, "A Guide to Letting Go of Outdated Google Search Best Practices".
How it works
Automated bidding is inherently different from manual bidding in the sense that the bidding algorithm can actually learn and make incremental improvements over time. "How does auto-bidding learn to make these improvements?" you ask? Simply put, the bidding algorithm takes in all the data points we feed in over time and learns to recognize patterns. These patterns allow the bidding algorithm to recognize patterns that will lead to success (such as a conversion) or patterns that will lead to failure (a non-converting click).
However, this means that in order to be successful with automated bidding, we need to structure our accounts to maximize the amount of data in a single bid strategy. According to Danis Post, the days of hyper-segmentation of our account structures are over. The goal is to maximize the number of impressions or clicks / ad groups. This way we can learn the automated tendering process as quickly as possible.
Trying to use automated bidding while following yesterday's best practices is likely one reason you may not have found success with automated bidding yet. I sure haven't found that automated bidding works right away, but through trial and error, I've found that automated bidding can not only outperform manual bids, but also save you a lot of time. Time that is better spent on strategic planning than manually changing bids once a week.
The key here is consolidation. Consolidate your campaigns / ad groups as much as possible and segment these new campaigns or ad groups based on intent or business goal instead of match type if possible.
After all, time is an important factor. It's foolish to turn on auto-bidding and expect improved performance overnight. It will likely take a few weeks for performance to improve sustainably. We are often under great pressure from customers to continually improve performance and / or to immediately try to "fix" the account when the performance deteriorates.
It is certainly possible to see a drop in performance at the beginning of an automated bid test as the algorithm learns. We account managers have to set this expectation with our customers before we start the test. Enter the test time. Do not leave the test immediately! I would recommend running all automated bid tests at least 2-3 weeks before drawing any conclusions.
Getting started with automated bidding is certainly a leap of confidence, but there are things we can do before we ever run an automated bid test that can set us up for success! First, we can restructure the account to intentionally consolidate our campaigns / ad groups and move away from hyper-segmented account structures. Second, we can set the expectation with our customers that this test will start off a bit rocky. There is a light at the end of the tunnel, however, and that light can certainly mean a huge improvement in performance and time savings.